Fourier-Mode Biot–Savart Velocity and Frequency Scaling

Good to see you again. In the previous lesson, we separated three claims in a smooth-forcing blowup result: the solution is smooth at every $t<T$, the forcing is smooth even at $T$, yet derivative norms of the solution can diverge as $t\uparrow T$. The basic route to that divergence is fine-scale structure: a field can have modest amplitude while its spatial derivatives become very large.

We now examine one such building block exactly. A single temperature–vorticity wave has a physical wavevector $q=\lambda \zeta$. Solving the two-dimensional Biot–Savart law for its velocity will show a crucial asymmetry: at fixed vorticity amplitude, the velocity of a high-frequency wave is smaller by one power of frequency, while the temperature gradient is larger by one power of frequency.

---

The local plane-wave building block

The construction first works in an ideal local affine setting, before spatial cutoffs are introduced. The old fields have the form

$$u_{\mathrm{old}}(x,t)=D(t)x,{\theta }_{\mathrm{old}}(x,t)=G(t)\cdot x,$$

where incompressibility requires

$$\mathrm{tr}D(t)=0.$$

A new layer is added as a temperature perturbation $\vartheta$ and a scalar-vorticity perturbation $\varpi$. At a fixed time, suppress the time dependence and set

$$q=\lambda \zeta ,K=|q|=\lambda |\zeta |.$$

Here $\lambda$ is the large frequency parameter used to separate successive layers, while $\zeta$ is a nonzero direction-and-scale vector that the background flow may transport.

The sinusoidal ansatz is

$$\vartheta (x)=\Theta \sin (q\cdot x),\varpi (x)=\Omega \cos (q\cdot x).$$

The signed constants $\Theta$ and $\Omega$ are the temperature and vorticity amplitudes. The apparent phase difference between $\vartheta$ and $\varpi$ is deliberate: taking a spatial derivative of sine produces cosine.

The original paper introduces this local wave before adding cutoffs:

Before calculating, fix one convention. For a two-dimensional velocity $v=(v_1,v_2)$, define scalar vorticity by

$$\varpi ={\partial }_1{v}_2-{\partial }_2{v}_1.$$

Introduce the streamfunction $\psi$ through

$$v={\nabla }^{\perp }\psi ,{\nabla }^{\perp }\psi =(-{\partial }_2\psi ,{\partial }_1\psi ).$$

Equivalently, define the counterclockwise quarter-turn matrix

$$J=\left(\begin{array}{cc}0& -1\\ 1& 0\end{array}\right),$$

so that

$${\nabla }^{\perp }\psi =J\nabla \psi .$$

This automatically gives $\nabla \cdot v=0$, and direct differentiation gives

$$\varpi =\Delta \psi .$$

Thus recovering velocity from vorticity amounts to solving a Poisson equation, followed by one spatial derivative.

---

Fourier-mode Biot–Savart law

For a general Fourier mode with nonzero wavevector $k$, write

$$\varpi (x)=\hat{\varpi }(k)e^{ik\cdot x}.$$

Since

$$\widehat{\Delta \psi }(k)=-|k{|}^2\hat{\psi }(k),$$

the Poisson equation $\Delta \psi =\varpi$ yields

$$\hat{\psi }(k)=-\frac{1}{|k{|}^2}\hat{\varpi }(k).$$

Applying $J\nabla$ in Fourier variables gives

$$\hat{v}(k)=iJk\hat{\psi }(k)=-\frac{iJk}{|k{|}^2}\hat{\varpi }(k).$$

That is the two-dimensional Fourier-space Biot–Savart law under our convention for $J$.

Many texts define the perpendicular vector in the clockwise direction,

$$k^{\perp }=(k_2,-k_1)=-Jk.$$

With that notation, the identical formula becomes

$$\hat{v}(k)=\frac{i{k}^{\perp }}{|k{|}^2}\hat{\varpi }(k).$$

The sign convention is not substantive. The central fact is the multiplier:

$$\left|\frac{k^{\perp }}{|k{|}^2}\right|=\frac{1}{|k|}.$$

Biot–Savart inversion is therefore an operator of order $-1$: velocity is one spatial derivative smoother than vorticity.

The following short passage gives the same general law in the periodic or whole-space setting.

There is an intuitive way to read this multiplier. To obtain $\psi$ from $\varpi$, solving the Poisson equation costs two powers of frequency:

$$\psi \text{has scale}\frac{\varpi }{K^2}.$$

Recovering velocity differentiates $\psi$ once, restoring one power:

$$v\text{has scale}\frac{\varpi }{K}.$$

Recovering the velocity gradient differentiates once more:

$$\nabla v\text{has scale}\varpi .$$

So velocity is smoother than vorticity, but its gradient is at the same frequency-scale order as vorticity.

---

Applying Biot–Savart to the cosine vorticity wave

Return to the particular wave

$$\varpi (x)=\Omega \cos (q\cdot x),|q|=K.$$

Since

$$\Delta \cos (q\cdot x)=-K^2\cos (q\cdot x),$$

a streamfunction that solves $\Delta \psi =\varpi$ is

$$\psi (x)=-\frac{\Omega }{K^2}\cos (q\cdot x).$$

Now differentiate:

$$\nabla \psi (x)=\frac{\Omega }{K^2}q\sin (q\cdot x).$$

Applying the perpendicular gradient gives the velocity wave

$$v(x)=\frac{\Omega }{K^2}Jq\sin (q\cdot x).$$

Using $q=\lambda \zeta$ and $K^2={\lambda }^2|\zeta {|}^2$, this is equivalently

$$v(x)=\frac{\Omega }{\lambda |\zeta {|}^2}J\zeta \sin (\lambda \zeta \cdot x).$$

This formula has three geometric features worth retaining.

First, $v$ is tangent to the wavefronts. The phase $q\cdot x$ is constant on lines perpendicular to $q$, while $Jq$ is perpendicular to $q$. Thus the velocity oscillates along, rather than across, those lines.

Second, the velocity has a sine phase, whereas vorticity has a cosine phase. At a vorticity maximum, such as a point where $q\cdot x=0$, this velocity perturbation vanishes.

Third, its amplitude is

$$\Vert v{\Vert }_{\mathrm{amp}}=\frac{|\Omega |}{K}=\frac{|\Omega |}{\lambda |\zeta |}.$$

Hence, if the vorticity amplitude $\Omega$ is held fixed while frequency increases, the velocity amplitude decays like $K^{-1}$.

This is a concrete manifestation of the order $-1$ Biot–Savart operator. A very fine vorticity oscillation need not carry a large velocity.

---

Temperature gradients: where frequency becomes expensive

The temperature wave is

$$\vartheta (x)=\Theta \sin (q\cdot x).$$

Taking one derivative gives

$$\nabla \vartheta (x)=\Theta q\cos (q\cdot x).$$

Therefore the temperature-gradient amplitude is

$$\Vert \nabla \vartheta {\Vert }_{\mathrm{amp}}=|\Theta |K=\lambda |\zeta ||\Theta |.$$

In particular, if $\Theta$ stays fixed and $|\zeta |$ remains comparable to one, then

$$\Vert \vartheta {\Vert }_{\mathrm{amp}}=|\Theta |,\Vert \nabla \vartheta {\Vert }_{\mathrm{amp}}\sim \lambda |\Theta |.$$

A short wave can therefore have bounded temperature amplitude but arbitrarily large temperature gradient.

For comparison, differentiating the velocity formula gives

$$\nabla v(x)=\frac{\Omega }{K^2}(Jq)\otimes q\cos (q\cdot x).$$

Writing $e=q/K$ for the unit wavevector direction, this becomes

$$\nabla v(x)=\Omega (Je)\otimes e\cos (q\cdot x).$$

At a phase maximum, for instance $q\cdot x=0$,

$$\nabla v(x)=\Omega (Je)\otimes e.$$

Its size is of order $|\Omega |$, independent of $K$ when $\Omega$ is fixed. This is the layer’s shear: displacement in the $e$ direction produces velocity in the perpendicular $Je$ direction.

The full frequency accounting is compactly summarized here.

Quantity	Plane-wave amplitude	Scaling at fixed $\Theta ,\Omega$ as $K$ increases
Temperature $\vartheta$	$	\Theta
Temperature gradient $\nabla \vartheta$	$K	\Theta
Vorticity $\varpi$	$	\Omega
Velocity $v$	$	\Omega
Velocity gradient $\nabla v$	$	\Omega

There are thus two different ways a construction can make a derivative large:

1. It can raise the wave frequency $K$ while retaining a non-negligible temperature amplitude $\Theta$, producing a large $\nabla \vartheta$.

2. It can raise the vorticity amplitude $\Omega$, producing a large $\nabla v$.

The Boussinesq coupling will connect these routes. A large horizontal component of $\nabla \vartheta$ sources vorticity, and the velocity induced by vorticity acts on the older temperature gradient. Later, the carefully chosen amplitude dynamics will relate $\Omega$ and $\Theta$, rather than treating them as unrelated fixed constants.

For now, be careful with one qualification. It is correct that high frequency makes velocity small at fixed $\Omega$:

$$\Vert v{\Vert }_{\mathrm{amp}}=\frac{|\Omega |}{K}.$$

But the eventual growing mode can itself make $\Omega$ depend on $K$. Thus “high frequency means small velocity” is not an unconditional statement; the exact conclusion is that Biot–Savart supplies one inverse power of frequency relative to vorticity.

---

Why this calculation matters for the blowup mechanism

The construction needs to accumulate increasingly large gradients without simply making every field amplitude enormous. This wave calculation supplies the necessary separation of scales.

A temperature wave with amplitude $\Theta$ contributes

$$K|\Theta |$$

to the size of its gradient. Thus frequency growth can create a large vorticity source even if the temperature contribution itself remains small in amplitude.

Meanwhile, its accompanying velocity is not comparably large merely because the wave is fine:

$$|v|\sim \frac{|\Omega |}{K}.$$

The high-frequency wave is dynamically potent through its derivatives, while its velocity amplitude is damped by Biot–Savart inversion. This distinction is one reason it is meaningful to target derivative blowup rather than amplitude blowup.

The orientation information is equally important. The temperature gradient is parallel to $q$, whereas the induced velocity is parallel to $Jq$. In the next lesson, that perpendicularity will explain why the shared-phase temperature and vorticity waves avoid nonlinear self-advection. That exact cancellation is what lets this elementary Fourier calculation become a useful nonlinear building block rather than only a linear heuristic.

---

Takeaways

For the plane-wave layer

$$\vartheta =\Theta \sin (\lambda \zeta \cdot x),\varpi =\Omega \cos (\lambda \zeta \cdot x),$$

the physical wavevector is

$$q=\lambda \zeta ,K=|q|=\lambda |\zeta |.$$

Solving $\Delta \psi =\varpi$ and setting $v={\nabla }^{\perp }\psi$ gives

$$\psi =-\frac{\Omega }{K^2}\cos (q\cdot x),$$ $$v=\frac{\Omega }{K^2}Jq\sin (q\cdot x).$$

Consequently,

$$\Vert \nabla \vartheta {\Vert }_{\mathrm{amp}}=K|\Theta |,\Vert v{\Vert }_{\mathrm{amp}}=\frac{|\Omega |}{K},\Vert \nabla v{\Vert }_{\mathrm{amp}}=|\Omega |.$$

The central conceptual result is that temperature gradients gain one power of frequency, while velocity loses one power relative to vorticity. Next, we use the perpendicular orientations of $v$, $\nabla \vartheta$, and $\nabla \varpi$ to show that these shared-phase waves do not advect themselves.